IX2-0439
① SA Source
- Source: 開啟完整 SA 文章
- Section:
WideEP - Line hint:
658
Context Before
Adding tensor parallelism within groups of GPUs addresses this. In an EP64/DP8/TP8 configuration, the 64 GPUs are organized into 8 DP groups of 8 GPUs each. Within each TP group, the attention projections, shared expert, normalization, and LM head are sharded 8 ways, so each GPU holds only 1/8th of the non-expert weights. Across the full cluster, the 256 experts are still distributed one-per-4-GPUs as before.
Pure DEP has a single communication pattern: all-to-all for expert routing. Adding TP introduces a second all-reduce within each TP group for the attention and non-expert computations. The key design principle is to place TP groups within a single node, where NVLink or MNNVL provides high-bandwidth interconnect, and run EP/DP across nodes, where the all-to-all communication pattern can tolerate higher latency.
Evidence
But each DP group now processes attention 8x faster per step, since the matmul is split 8 ways across the TP group
Context After
Disaggregated Prefill
Disaggregated prefill, sometimes referred to as prefill-decode (PD) disaggregation, is the process of performing prefill and decode phases of LLM inference on separate nodes. Prefill occurs when a request is first processed, and a forward pass is computed on all tokens at once, thereby “prefilling” the KV cache for this request. This is a compute-intensive operation as all tokens feed through the forward pass in parallel. Tokens are then generated or “decoded” one at a time, loading the KV cache from HBM at each decode step. This is a memory-intensive process as the growing KV cache is constantly being loaded.
② Atomic Claim
但每個 DP group 的 attention step 會快約 8 倍，因為 matmul 被分散到 TP group 中 8 顆 GPU。
- Epistemic Mode:
ASSERTED - Mapping Status:
COMPLETE
③ Semantic Frame
{
"comparison_expression": "但每個 DP group 的 attention step 會快約 8 倍，因為 matmul 被分散到 TP group 中 8 顆 GPU。",
"entities": [
{
"id": "04_knowledge_base/Tensor Parallelism",
"label": "TP"
},
{
"id": "04_knowledge_base/GPU",
"label": "GPU"
}
],
"frame_type": "COMPARISON",
"metric": "COUNT",
"operator": "MULTIPLE_OF",
"qualifiers": {
"condition_text": null,
"numeric_mentions": [
"8"
],
"temporal_mentions": []
}
}④ Canonical Entity Mapping
| Role | Surface Label | Canonical Target |
|---|---|---|
| comparison_entity_0 | TP | 04_knowledge_base/Tensor Parallelism |
| comparison_entity_1 | GPU | GPU |
⑤ Human Review
請在 Properties 逐項確認：
- 原文 → Atomic Claim 是否忠實
- Atomic Claim → Semantic Frame 是否忠實
- Canonical Entity mapping 是否正確
- Epistemic mode 是否保留原文語氣
- 最後選擇
review_action
Review state
Markdown 內文不是正式 approval。只有 Apply bridge 寫入的 Decision Ledger event 才是正式決策。